Comment on ‘Optimal weekly releases from a seasonal reservoir: 1. Deterministic future’ by F. Laufer and H. J. Morel‐Seytoux

In a large class of problems the Kuhn-Tucker optimality conditions [Hillier and Liebertnan, 1972; This information is not always constructive, however, as far as the actual derivation of the optimal solutions is concerned, since although in most cases it is a fairly simple matter to derive these conditions, it is often quite difficult to solve them. Moreover, in general, they constitute only necessary optimality conditions, and unless certain convexity conditions are also satisfied there is no guarantee that the solutions they generate are optimal. For this reason the use of these conditions as a basis for solution algorithms has traditionally been restricted to problems satisfying certain separability and convexity conditions [Hillier and Liebertnan, 1972;